EXECUTIVE SUMMARY
FROM THE CROSS-SITE AT HOME/CHEZ SOI PROJECT
Homelessness is a serious public policy concern.
Each year, up to 200,000 people are homeless in
Canada — at an estimated cost of seven billion dollars.
In Canada, our current response relies heavily upon shelters for emergency housing and emergency and crisis services for health care. Typically, individuals who are homeless must first participate in treatment and attain a period of sobriety before they are offered housing. This is a costly and ineffective way of responding to the problem.
Alternatively, Housing First (HF) is an evidence-based intervention model, originating in New York City (Pathways to Housing), that involves the immediate provision
of permanent housing and wrap-around supports to individuals who are homeless and living with serious mental illness, rather than traditional “treatment then housing”
approaches. HF has been shown to improve residential stability and other outcomes. Given the difference in social policy and health care delivery between the U.S.
and Canada, it is vital that evidence about homelessness interventions be grounded in the Canadian context.
In 2008, the federal government invested $110 million for a five-year research demonstration project aimed at generating knowledge about effective approaches for people experiencing serious mental illness and homelessness in Canada. In response, the Mental Health Commission of Canada (MHCC) and groups of stakeholders in five cities (Vancouver, Winnipeg, Toronto, Montréal, and Moncton) implemented a pragmatic, randomized controlled field trial of HF. The project, called At Home/Chez Soi, was designed to help identify what works, at what cost, for whom, and in which environments. It compared HF with
existing approaches in each city. The examination of quality of life, community functioning, recovery, employment, and related outcomes was unprecedented, as was the inclusion of two types of support services for individuals with high needs (Assertive Community Treatment, or ACT) and moderate needs (Intensive Case Management, or ICM).
The study also used a standardized model of HF, conducted assessments of program fidelity to document the quality of program implementation, introduced quality assurance processes, and provided extensive training, technical assistance, and support.
A randomized trial design was used in the project because it could evaluate the effects of HF in groups that were virtually identical except for the intervention itself, thus giving the strongest evidence for policy. The study also included a qualitative research component to complement and better inform the quantitative results (mixed methods design). Data collection began in October 2009 and ended in June 2013.
2,148 individuals were enrolled for two years of follow-up and of those, 1,158 received the HF intervention. Follow-up rates at 24 months were between 77 and 89 per cent, which are excellent for a vulnerable and highly transient population.
This document reports on the main findings of the study for the full two years of follow-up. It builds on the At Home/
Chez Soi Interim Report (September 2012), which presented the preliminary one-year results. Reports containing greater detail about local findings and implications for local practice and policy are also available for each of the five cities.
Homelessness is a serious public policy concern.
Each year, up to 200,000 people are homeless in
Canada — at an estimated cost of seven billion dollars.
Program Implementation
The study demonstrated that HF can be implemented successfully in different Canadian contexts, using both ACT and ICM models for the service component.
It also demonstrated that HF can be effectively adapted according to local needs, including rural and smaller city settings such as Moncton and communities with diverse mixes of people (e.g., Aboriginal or immigrant populations) like Winnipeg or Toronto.
Study Participants
Most At Home/Chez Soi study participants were recruited from shelters or the streets.
The typical participant was a male in his early 40s, but there was a wide diversity of demographic characteristics. Women (32 per cent), Aboriginal people (22 per cent), and other ethnic groups (25 per cent) were well-represented. The typical total time participants experienced homelessness in their lifetimes was nearly five years. Participants were found to have had multiple challenges in their lives that contributed to their disadvantaged status.
For example, 56 per cent did not complete high school, and almost everyone was living in extreme poverty at study entry.
All had one or more serious mental illness, in keeping with the eligibility criteria of the study, and more than 90 per cent had at least one chronic physical health problem. Using qualitative interviews with a representative sample and quantitative measures, we have documented the early origins of homelessness in the life histories of participants, which very often included early childhood trauma and leaving home to escape abuse.
Housing Outcomes
HF was found to have a large and significant impact on housing stability.
A substantial majority of participants maintained stable housing during the study period, indicating that the attention paid to client choice and service team support quickly resulted in securing desirable and affordable housing. In the last six months of the study, 62 per cent of HF participants were housed all of the time, 22 per cent some of the time, and 16 per cent none of the time; whereas 31 per cent of treatment as usual (TAU)
participants were housed all of the time, 23 per cent some of the time, and 46 per cent none of the time. These significant gains in obtaining and retaining housing held for participants in both the ACT and ICM versions of HF. Over the course of the study, TAU participants spent significantly more time in temporary housing, shelters, and on the street than HF participants.
The most dramatic effects were found in the first year, where the HF program
“jumpstarts” getting housed. Many HF participants spoke of the importance of
“having their own place” and described their housing as a safe and secure “base”
from which to move forward with their lives. One noted, “The security is a really big thing. I can just let go and I have no problem just lying down for 12 hours and I don’t have to move or be on guard.”
(Vancouver participant)
Clients with Additional or Other Needs
HF worked well for clients with diverse ethnocultural backgrounds and
circumstances. We now know more about the small group (about 13 per cent) for whom HF as currently delivered did not result in stable housing in the first year.
This group tended to have longer histories of homelessness, lower educational levels, more connection to street-based social networks, more serious mental health conditions, and some indication of greater cognitive impairment. Alternative approaches to addressing the unique needs of these clients were tried in some cities. Recommendations on these approaches will be available in the Housing First implementation toolkit.
Housing quality
Our field research teams systematically measured housing quality using standard ratings in a random sample of 205 HF and 229 TAU residences. The HF residences (unit and building combined) were found across sites to be of significantly greater quality and of much more consistent quality than those that TAU participants were able to get on their own or using other housing programs and services.
There were moderate site differences in these findings.
Costs and Service Use
One of the advantages of stable housing for a group who have high levels of chronic mental and physical illness is the possibility of shifting their care from institutions to the community. Community services including visits from the HF service providers and phone contacts increased as intended and, particularly for the high needs group, inpatient and crisis- type service use fell. Most of the service use changes reflect appropriate shifts from crisis services to community services, but for some participants, involvement in the program likely resulted in the identification of unmet needs for more acute or rehabilitative levels of care in the short term. These shifts in service use create cost savings and cost offsets that can be taken into account when making decisions about where to target future programs and how to avoid future cost pressures.
The economic impact of HF was also studied, considering all costs incurred by society. HF cost $22,257 per person per year on average for high needs participants, and $14,177 per person per year for moderate needs participants.
Program costs include staff salaries and expenses such as travel, utilities, and rent supplements. HF for high needs participants is more costly mainly because of the higher staff:participant ratio. Over the two-year period after study entry, HF services resulted in average reductions of $21,375 in service costs for high needs participants, and $4,849 for moderate needs participants. Thus, every $10 invested in HF resulted in an average savings of $9.60 for high needs participants and $3.42 for moderate needs participants. This net savings arises from a combination of decreases in the costs of some services (cost offsets), and increases in the costs of others. For high needs participants, the main cost offsets were psychiatric hospital stays, home and office visits to health or social service providers, and jail or prison stays. For moderate needs participants, the main cost offsets were shelter stays and stays in single room accommodations with support services. For moderate needs participants, cost increases were seen in general hospital stays in psychiatric units.
For the 10 per cent of participants with the highest service use costs at the start
Living in shelters and on the streets requires that enormous energy be put into basic survival. The circumstances are not conducive to participating in treatment and managing health issues. On average, participants had been homeless in their lifetime for just less than five years when they enrolled in the study, and many had a history of poverty and disadvantage reaching back to early childhood. For some, the road to recovery after housing can be rapid, but for most it is more gradual and setbacks are to be expected.
In general, the study documented clear and immediate improvements, followed by more modest continuing ones for the remainder of the study period. Some outcomes, including mental health and substance use problems, improved by a similar amount in both HF and TAU. These improvements may be due to services that can be accessed by both groups, or may represent natural improvement after a period of acute homelessness. However,
gains in participant-reported quality of life and observer-rated community functioning were significantly greater in HF (for both ACT and ICM) than in TAU. These differences were relatively modest, but still represent meaningful improvement in outcomes for HF
compared to existing services, and indicate that HF can impact broader outcomes.
One Toronto participant described their experience as: “I am really proud of myself, with a lot of help I was…able to…not really get back to where I used to be, but in a better place.” (Toronto participant) While the HF groups on average improved more on the major outcomes, the individual responses in both HF (ICM and ACT) and TAU over time were enormously diverse.
Across all sites in the qualitative interviews, 61 per cent of the HF participants described a positive life course since the study began, 31 per cent reported a mixed life course, and eight per cent reported a negative life course. In contrast, only 28 per cent of
TAU reported a positive life course, 36 per cent reported a mixed life course, and 36 per cent reported a negative life course.
The study generated and consolidated rich information about different sub- populations, diverse responses, and how to successfully adapt the approach.
Housing stability, quality of life, and community functioning outcomes were all more positive for programs that operated most closely to HF standards, including the provision of rent subsidies. HF model standards were measured on 38 items in five domains for 12 programs at two time points in the study (early implementation and one year later). Overall there was strong fidelity to HF standards (with all items rated above 3 on a 4-point scale), and this improved over time (71 per cent in round one and 78 per cent in round two). This indicates that supporting all components of the HF model and investing in training and technical support can pay off in improved outcomes.
“I am really proud of myself, with a lot of help I was…
able to…not really get back to where I used to be, but in a better place.” (Toronto participant)
of the study, HF cost $19,582 per person per year on average. Receipt of HF services resulted in average reductions of $42,536 in the cost of services compared to usual care participants. Thus every $10 invested in HF services resulted in an average savings of $21.72. The main cost offsets were psychiatric hospital stays, general hospital stays (medical units), home and office visits with community-based providers, jail/prison incarcerations, police contacts, emergency room visits, and stays in crisis housing settings and in single room accommodations with support services. For this group, two costs increased: hospitalization in psychiatric units in general hospitals and stays in psychiatric rehabilitation residential programs.
Quality of Life, Functioning, Mental Health, and Substance Use Outcomes
